NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KBH37WVJM0Z4P3DB4NQYT6
Run Details
Model:
DeepSeek V4 Pro 0813 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
1m 2s
Cost:
$0.009006
Provider:
DigitalOcean
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0000%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 30% words of dialogue between two thieves named Molly and Bill who are planning to break into a children's party to steal all the presents. Only include the prose of the scene without any markdown or commentary.
Model response
Messages
Meta
AI
Molly had the binoculars trained on the bouncy castle, which was currently absorbing the kinetic energy of seven sugar-crazed six-year-olds. Bill sat beside her in the van, smoothing a crinkled napkin blueprint of the backyard party layout with the reverence of a museum curator. On the napkin, he had drawn a cake with a skull and crossbones. “Phase one,” Bill whispered, tapping the napkin. “We infiltrate under the cover of the piñata.” “The piñata is shaped like a unicorn, Bill.” “Exactly. No one suspects the unicorn.” Molly lowered the binoculars and stared at him. “We’re wearing clown shoes and scratchy wigs. The five-year-olds will know we’re not clowns because we won’t be crying.” Bill shook his head. “Clowns don’t cry. They honk. I’ve practiced.” He pulled a tiny bicycle horn from his pocket and honked twice. A passing dog barked. Molly slapped the horn out of his hand. “Focus. The gift table is under the inflatable rainbow arch. Present pile. Right side.” She gestured with her chin, but Bill was drawing a second skull and crossbones on the napkin, this one with a party hat. “Phase two: the distraction.” Bill’s voice dropped to a stage whisper. “I release the ferret.” “We don’t have a ferret.” “I’ve improvised. It’s a sock puppet with googly eyes. Very convincing.” Molly massaged her temples. “A sock puppet is not a distraction. The real distraction is you getting your clown wig caught in the cotton candy machine. Which will happen. We both know it.” Bill beamed. “That’s the beauty of it. While I’m being rescued from the cotton candy machine, you sneak through the present pile, stuff the bags with plush toys and glitter gift cards, and we meet at the bouncy castle.” “The bouncy castle is full of children.” “They’ll clear out when they see you. You have the face.” “What face?” “The face that says ‘this adult is here to steal my birthday Barbie.’” Bill snapped the napkin blueprint into a tiny paper hat and put it on his head. “It’s a classic misdirection.” Molly sighed so hard the van windows fogged. “We are not doing the clown shoes. We agreed. No costumes. Just the vests.” “The vests say ‘Party Crew’ on the back. It’s a step down from clown.” “It’s a step up from a sock puppet ferret.” Bill held up the sock puppet. It had googly eyes and a tiny red tongue made of felt. “He’s named Reginald.” “He’s a tube sock.” “Reginald has a degree in misdirection.” Molly grabbed the sock puppet and tossed it into the back of the van, where it landed in a pile of crumpled fast-food wrappers. “Here’s the real plan. We wear the vests. We walk in. We fill two insulated coolers with ice and juice boxes. We hide the presents under the cake boxes. We walk out. Nobody asks because it looks like we’re cleaning up. It’s a birthday party, Bill. Everyone leaves with a paper plate and a sense of delayed gratitude.” Bill crossed his arms. “That’s boring.” “It’s foolproof. Boring is the best thing a heist can be.” She checked her watch. “They sing happy birthday in twelve minutes. Cake cutting causes a crush of parents with phones. That’s our window.” Bill looked genuinely wounded. “So Reginald doesn’t get a moment?” “No.” “Not even a honk?” “No honks. No ferrets. No unicorn piñata infiltration. We are two stressed caterers with a plausible reason to carry a dozen wrapped gifts out the side gate.” Molly reached into the glovebox and pulled out two clipboards. She handed one to Bill. “Here. If anyone asks, we’re doing a headcount for the fire marshal.” Bill read the clipboard upside down. “It says ‘Gift Inventory: Bikes, Legos, gift cards, one very heavy box that is probably a drum set we should avoid at all costs.’” “Good. You’re learning.” “I still think the sock puppet would’ve added texture.” Molly opened the van door and stepped into the sticky heat of a Sunday afternoon. Somewhere beyond the fence, a child screamed with joy. Bill followed, adjusting his vest, and for one beautiful second, he did not honk the horn. He simply whispered, “Reginald sends his regards.”
created_at
"2026-09-03T10:02:45.91Z"
9/3/2026, 10:02:45 AM
model
"deepseek/deepseek-v4-pro-20260813"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
508
moderation_latency
(null)
generation_time
62122
tokens_prompt
75
tokens_completion
2480
native_tokens_prompt
67
native_tokens_completion
2252
native_tokens_completion_images
(null)
native_tokens_reasoning
1187
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
num_fetches
(null)
web_search_engine
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
service_tier
(null)
usage
0.00900636
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788429765-lkRAJl6MUguIM6GmVahY"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788429765-By4iL1Ha3BY1Cl3wm3mk"
upstream_id
"chatcmpl-359e0cb3-ede5-4edb-a579-5314f1c1e766"
provider_responses
0
endpoint_id
"928ab2e9-a85f-427f-adcb-15b2394c70d5"
id
"chatcmpl-359e0cb3-ede5-4edb-a579-5314f1c1e766"
is_byok
false
latency
507
model_permaslug
"deepseek/deepseek-v4-pro-20260813"
provider_name
"DigitalOcean"
status
200
total_cost
0.00900636
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DigitalOcean"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
697
0.00%
Dialogue to Total Word Ratio
Ratio: 60.17%, Deviation: 30.17%
neededClean
false
wordsTotal
703
wordsDialogue
423
0.0000%